Some stats about Bitcoin multisig
More than a year ago, I published a state of Bitcoin multisig, presenting its history and its uses. Since then, the number of bitcoins stored in multisig setups has been relatively similar, hovering between 9 and 11% of all bitcoins (around $1.3B now). However, with a complete rewrite of p2sh.info’s backend, I am now able to tell you more about how multisig is used and what can we learn from it.
Introduction
Multisignature transactions in Bitcoin can range from 1 signature required out of 1 possible key (1-of-1) up to 15-of-15 keys with all the variants in between like 5-of-9, etc.. Interestingly, out of the 120 possible combinations, only 31 have been used so far, and only 13 setups are currently holding more than 100 BTC.
The most popular (in number of transactions) is 2-of-3, very commonly used in wallets like Bitgo, Copay, Greenaddress and more.
Looking at when multisig outputs are created and spent, it’s possible to reconstruct how much BTC is stored in each multisig setup for every point in time. This is what I have done and I’m going to show some of my findings
BTC and UTXO repartition
While 2-of-3 is by far the most commonly used multisig setup, in term of value stored, 3-of-5 has been the dominant one for most of multisig’s history. Currently, at least 585k BTC (~$480M) are stored in such addresses, but there has been as many as 930kBTC secured by this setup. Coinbase’s multisig vaults, Copay and others allow creation of 3-of-5 addresses.
At the other end of the scale, 1-of-10 has only been used once and currently stores 1 satoshi.
Here’s how bitcoins stored in P2SH multisig have been split among the different multisig setup over time:
For most of its history (March 2012 to late 2014), P2SH wasn’t used to store large amounts of BTC, however, with the advent of tools like Bitgo Entreprise, Copay, etc.., this quickly changed. Very quickly, 3-of-5 multisig(the blue one) imposed itself as the most BTC-rich, storing more than 0.93M BTC at one point.
The effect of Bitfinex’s hack in early August 2016 can also clearly be seen on the 2-of-3 setup (in yellow) with this setup losing nearly half of the BTC it stored in one week. Even to this point, 2-of-3 multisig never really recovered from the hack, never breaking its previous ATH in BTC stored.
The drop in BTC stored in all setups at the end of the chart is artificial. It is due to BTC moving from addresses with a known setup to ones that haven’t spent any BTC yet, thus not revealing their underlying BTC setup.
As for the UTXO set (unspent transaction outputs), it has also been steadily growing, this time, 2-of-3 multisig represents the bulk of the UTXO set, with 2-of-2 multisig second and 3-of-5 third.
Whale spotting
Looking at how the repartition of BTC changed between different multisig setups makes it possible to notice changes in policy of entities storing large amounts of BTC (most probably exchanges).
For example, 5 days after Bitfinex’s hack in August 2016 (the first drop in the yellow line), an entity moved around 75k BTC from a 2-of-3 setup (which was used by Bitfinex) to a 3-of-5 one (the green line).
Entities being the only ones to use distinctive setups expose their way of using their bitcoins. For example, between mid July 2016 and the end of September of the same year, an entity used P2SH addresses with 3-of-6 multisig to store 37k BTC and then spent 17k BTC progressively over that period.
Conclusion
Ever since I began charting the Bitcoin blockchain nearly 3 years ago, I’ve always found it fascinating that you could access all the transactional data of a billion-scale financial network and gain insights into its actors’ behaviours just by charting it.
I’m sure this new data has plenty more secrets to reveal, that’s why I’m working on making it publicly available on p2sh.info, in a live-updated version.
If you enjoyed this article, do not hesitate to contribute to my beer/hosting fund:
